The ML&DS Lab aims to design algorithms and to develop tools for getting insights from large amounts of data, in particular, social media, user behavior, security and bioinformatics data. Our current projects are focused on:

• Domain Adaptation Approaches for Classifying Crisis-related Social Media Data
• Domain Adaptation and Semi-supervised Approaches for Biological Sequence Classification
• Semi-supervised Approaches for Android Malware Detection
• Cross-domain Approaches for Recommender Systems

The common denominator for these projects is the limited amount of labeled data for a problem of interest. However, large amounts of unlabeled data are readily available. Furthermore, labeled data for a different, but related problem might also be available. To handle the scarcity of labeled data and employ the available unlabeled data, we focus on the design of semi-supervised and domain adaptation approaches, and use our approaches to learn classifiers for a variety of classification problems. More information on specific projects is available at projects' website.